问题
无论我如何到达这里，我真正想做的就是完全删除 nvidia 和 cuda 驱动程序并重新安装它们。但是，无论我做什么，我都无法摆脱它们。
背景
我在 Ubuntu 22.04 上，我使用一个有效的 nvidia-535 驱动程序，并成功安装了 cuda 11-8（使用此处的说明）。
现在，由于某些（对我来说）未知的原因（可能是某些更新），cuda 停止工作（即稳定扩散中未使用 gpu）并nvidia-smi返回
Failed to initialize NVML: Driver/library version mismatch
NVML library version: 535.161
因此想尝试删除与 cuda 和 nvidia 相关的所有内容并重新开始，因为我知道它之前运行良好。
我努力了
sudo apt remove --purge nvidia*返回
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Note, selecting 'nvidia-driver-550-server' for glob 'nvidia*'
Note, selecting 'nvidia-firmware-535-535.154.05' for glob 'nvidia*'
Note, selecting 'nvidia-docker2' for glob 'nvidia*'
... [ca. 300 similar rows] ...
Package 'nvidia-egl-wayland-common' is not installed, so not removed
Note, selecting 'nvidia-settings' instead of 'nvidia-settings-binary'
Package 'nvidia-390' is not installed, so not removed
Note, selecting 'libnvtt-bin' instead of 'nvidia-texture-tools'
Package 'nvidia-libopencl1-dev' is not installed, so not removed
Package 'nvidia-current' is not installed, so not removed
Package 'nvidia-current-updates' is not installed, so not removed
Package 'nvidia-libopencl1' is not installed, so not removed
Package 'nvidia' is not installed, so not removed
... [ca. 500 similar rows] ...
Package 'nvidia-container-toolkit-base' is not installed, so not removed
Package 'nvidia-firmware-545-545.29.02' is not installed, so not removed
Package 'nvidia-firmware-550-550.40.07' is not installed, so not removed
You might want to run 'apt --fix-broken install' to correct these.
The following packages have unmet dependencies:
cuda-drivers-535 : Depends: nvidia-compute-utils-535 (>= 535.129.03) but it is not going to be installed
Depends: nvidia-dkms-535 (>= 535.129.03)
Depends: nvidia-driver-535 (>= 535.129.03) but it is not going to be installed
Depends: nvidia-kernel-common-535 (>= 535.129.03) but it is not going to be installed
Depends: nvidia-kernel-source-535 (>= 535.129.03) but it is not going to be installed or
nvidia-kernel-open-535 (>= 535.129.03) but it is not going to be installed
Depends: nvidia-utils-535 (>= 535.129.03) but it is not going to be installed
Depends: nvidia-modprobe (>= 535.129.03) but it is not going to be installed
Depends: nvidia-settings (>= 535.129.03) but it is not going to be installed
E: Unmet dependencies. Try 'apt --fix-broken install' with no packages (or specify a solution).
- 根据建议我尝试了：
sudo apt --fix-broken install返回：
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Correcting dependencies... Done
The following packages were automatically installed and are no longer required:
libsbc1 nvidia-firmware-535-535.146.02 nvidia-firmware-535-535.161.07
Use 'sudo apt autoremove' to remove them.
The following additional packages will be installed:
nvidia-kernel-common-535
The following packages will be upgraded:
nvidia-kernel-common-535
1 upgraded, 0 newly installed, 0 to remove and 28 not upgraded.
2 not fully installed or removed.
Need to get 0 B/38,3 MB of archives.
After this operation, 45,1 kB of additional disk space will be used.
Do you want to continue? [Y/n] y
(Reading database ... 256847 files and directories currently installed.)
Preparing to unpack .../nvidia-kernel-common-535_535.161.07-0ubuntu1_amd64.deb ...
Unpacking nvidia-kernel-common-535 (535.161.07-0ubuntu1) over (535.154.05-0ubuntu1) ...
dpkg: error processing archive /var/cache/apt/archives/nvidia-kernel-common-535_535.161.07-0ubuntu1_amd64.deb (--unpack)
:
trying to overwrite '/lib/firmware/nvidia/535.161.07/gsp_ga10x.bin', which is also in package nvidia-firmware-535-535.1
61.07 535.161.07-0ubuntu0.22.04.1
dpkg-deb: error: paste subprocess was killed by signal (Broken pipe)
Errors were encountered while processing:
/var/cache/apt/archives/nvidia-kernel-common-535_535.161.07-0ubuntu1_amd64.deb
E: Sub-process /usr/bin/dpkg returned an error code (1)
sudo apt autoremove给出：
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
You might want to run 'apt --fix-broken install' to correct these.
The following packages have unmet dependencies:
nvidia-dkms-535 : Depends: nvidia-kernel-common-535 (= 535.161.07-0ubuntu1) but 535.154.05-0ubuntu1 is installed
E: Unmet dependencies. Try 'apt --fix-broken install' with no packages (or specify a solution).
sudo apt remove --purge "*cuda*"返回：
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Note, selecting 'libcuda-10.2-1' for glob '*cuda*'
Note, selecting 'libcudart.so.12' for glob '*cuda*'
... [ ca. 800 rows ] ...
Package 'libcudss0-cuda-12' is not installed, so not removed
Package 'libcudss0-dev-cuda-12' is not installed, so not removed
Package 'libcudss0-static-cuda-12' is not installed, so not removed
You might want to run 'apt --fix-broken install' to correct these.
The following packages have unmet dependencies:
libcublas-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libcufft-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libcufile-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libcurand-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libcusolver-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libcusparse-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libnpp-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libnvjpeg-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
nvidia-dkms-535 : Depends: nvidia-kernel-common-535 (= 535.161.07-0ubuntu1) but 535.154.05-0ubuntu1 is to be installed
E: Unmet dependencies. Try 'apt --fix-broken install' with no packages (or specify a solution).
正如另一篇文章中所述问题，我试图找到卸载程序：
sudo /usr/local/cuda-11.4/bin/cuda-uninstaller但下没有 cuda 或 nvidia 目录/usr/local/。我检查过
/etc/apt/sources.list，没有“nvidia”或“cuda”的实例（即cat /etc/apt/sources.list | grep nvidia没有返回任何内容）。按照附加说明进行操作问题，
sudo apt-get --purge remove "*cublas*" "cuda*" "nsight*"回到：
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Note, selecting 'cublasmp-cuda-11' for glob '*cublas*'
Note, selecting 'cublasmp-cuda-12' for glob '*cublas*'
... [ >100 similar rows ] ...
Package 'libcublas-dev-12-4' is not installed, so not removed
Package 'nsight-compute-2024.1.0' is not installed, so not removed
Package 'nsight-systems-2023.4.4' is not installed, so not removed
You might want to run 'apt --fix-broken install' to correct these.
The following packages have unmet dependencies:
libcufft-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libcufile-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libcurand-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libcusolver-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libcusparse-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libnpp-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
libnvjpeg-11-8 : Depends: cuda-toolkit-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-config-common but it is not going to be installed or
cuda-cudart-11-2 but it is not installable
Depends: cuda-toolkit-11-8-config-common but it is not going to be installed
nvidia-dkms-535 : Depends: nvidia-kernel-common-535 (= 535.161.07-0ubuntu1) but 535.154.05-0ubuntu1 is to be installed
E: Unmet dependencies. Try 'apt --fix-broken install' with no packages (or specify a solution).
总之，无论我尝试删除什么，它都不会让我删除，因为依赖关系尚未满足。我尝试逐个删除上面的最后一个软件包（例如 libcufft-11-8），但我再次得到非常相似的输出，列出了其他相关软件包。
另外，
sudo apt update返回：
Hit:1 https://download.docker.com/linux/ubuntu focal InRelease
Hit:2 http://se.archive.ubuntu.com/ubuntu jammy InRelease
Hit:3 http://se.archive.ubuntu.com/ubuntu jammy-updates InRelease
Hit:4 http://se.archive.ubuntu.com/ubuntu jammy-backports InRelease
Hit:5 https://dl.google.com/linux/chrome/deb stable InRelease
Hit:6 https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64 InRelease
Get:7 http://security.ubuntu.com/ubuntu jammy-security InRelease [110 kB]
Hit:8 https://ppa.launchpadcontent.net/graphics-drivers/ppa/ubuntu jammy InRelease
Hit:9 https://ppa.launchpadcontent.net/qbittorrent-team/qbittorrent-stable/ubuntu jammy InRelease
Get:10 http://security.ubuntu.com/ubuntu jammy-security/universe amd64 Packages [848 kB]
Get:11 http://security.ubuntu.com/ubuntu jammy-security/universe i386 Packages [597 kB]
Get:12 http://security.ubuntu.com/ubuntu jammy-security/universe Translation-en [162 kB]
Fetched 1 716 kB in 2s (1 033 kB/s)
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
29 packages can be upgraded. Run 'apt list --upgradable' to see them.
W: https://download.docker.com/linux/ubuntu/dists/focal/InRelease: Key is stored in legacy trusted.gpg keyring (/etc/apt/trusted.gpg), see the DEPRECATION section in apt-key(8) for details.
W: https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/InRelease: Key is stored in legacy trusted.gpg keyring (/etc/apt/trusted.gpg), see the DEPRECATION section in apt-key(8) for details.
并且sudo apt upgrade不起作用：
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
You might want to run 'apt --fix-broken install' to correct these.
The following packages have unmet dependencies:
nvidia-dkms-535 : Depends: nvidia-kernel-common-535 (= 535.161.07-0ubuntu1) but 535.154.05-0ubuntu1 is installed
E: Unmet dependencies. Try 'apt --fix-broken install' with no packages (or specify a solution).
系统详细信息
Ubuntu 22.04.4 LTS
内核：6.5.0-21-generic
答案1
我运行时刚刚遇到了这个问题sudo apt --fix-broken install
尝试运行类似这样的程序来修复它
sudo dpkg --force-all -i /var/cache/apt/archives/nvidia-kernel-common-535_535.161.07-0ubuntu1_amd64.deb
sudo apt --fix-broken install
然后重新启动机器。
答案2
一步步：
sudo apt-get purge nvidia*
sudo apt remove nvidia-*
sudo rm /etc/apt/sources.list.d/cuda*
sudo apt-get autoremove && sudo apt-get autoclean
sudo rm -rf /usr/local/cuda*
如果上述命令没有帮助，请通过 Grab 菜单启动并重试。
PS 在多个 Ubuntu 版本上都有效
答案3
在“软件和更新”中将驱动程序手动设置为 Nouveau 并重新启动后，、、sudo apt remove --purge nvidia*等sudo apt remove --purge "*cuda*"（如我的问题中所列以及 Siarhei 所建议的）按预期工作并且没有出现任何错误。
我知道这很简单，但以前（以及在其他系统上）我能够做到这一点，而无需手动激活 Nouveau 驱动程序并重新启动，所以我想当时我并没有想到这一点。
（感谢大家重新打开并回答问题。另外，对于打字错误，我深表歉意。）